Problem 034

Sum of numbers equal to the sum of the factorial of their digits.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 034
# Sum of numbers equal to the sum of the factorial of their digits.

function main() -> i32 {
    let fact: array<i64, 10> = [1, 1, 2, 6, 24, 120, 720, 5040, 40320, 362880]
    # Upper bound: 7*9! = 2540160
    let mut total: i64 = 0
    let mut n: i64 = 3
    while n <= 2540160 {
        let mut s: i64 = 0
        let mut x: i64 = n
        while x > 0 {
            s = s + fact[(x % 10) as i32]
            x = x / 10
        }
        if s == n {
            total = total + n
        }
        n = n + 1
    }
    printf("%lld\n", total)
    return 0
}

Generated C

#include <stdint.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

/* Flow runtime helpers */
typedef struct flow_temp_node { struct flow_temp_node* next; } flow_temp_node;
static flow_temp_node* flow_temp_head = NULL;
static int flow_temp_atexit_set = 0;
__attribute__((unused)) static void flow_temp_free_all(void) {
    while (flow_temp_head) {
        flow_temp_node* n = flow_temp_head;
        flow_temp_head = n->next;
        free(n);
    }
}
__attribute__((unused)) static void* flow_temp_alloc(size_t nbytes) {
    flow_temp_node* node = (flow_temp_node*)malloc(sizeof(flow_temp_node) + nbytes);
    if (!node) return NULL;
    node->next = flow_temp_head;
    flow_temp_head = node;
    if (!flow_temp_atexit_set) {
        flow_temp_atexit_set = 1;
        atexit(flow_temp_free_all);
    }
    return (void*)(node + 1);
}
#ifndef FLOW_DIAG
#define FLOW_DIAG(msg) fprintf(stderr, "%s", (msg))
#endif
#ifndef FLOW_LOG
#define FLOW_LOG(fmt, ...) printf(fmt, __VA_ARGS__)
#endif
#ifndef FLOW_LOG_EMPTY
#define FLOW_LOG_EMPTY(fmt) printf(fmt)
#endif
static char* flow_strcat(const char* a, const char* b) {
    size_t la = strlen(a ? a : ""), lb = strlen(b ? b : "");
    char* r = (char*)flow_temp_alloc(la + lb + 1);
    if (!r) return NULL;
    if (la) memcpy(r, a, la);
    if (lb) memcpy(r + la, b, lb);
    r[la + lb] = '\0';
    return r;
}

#define __flow_in_arr(arr, val) __extension__ ({ \
    int _found = 0; \
    size_t _n = sizeof(arr)/sizeof((arr)[0]); \
    for (size_t _i = 0; _i < _n; _i++) { \
        if ((arr)[_i] == (val)) { _found = 1; break; } \
    } _found; })

/* Unified fault handler (MISRA #279) — override with -DFLOW_FAULT_HANDLER=fn */
#ifndef FLOW_FAULT_HANDLER
__attribute__((unused)) static inline void flow_fault_handler(const char* msg) {
    fprintf(stderr, "flow: %s\n", msg ? msg : "fault");
    abort();
#if defined(__GNUC__) || defined(__clang__)
    __builtin_unreachable();
#endif
}
#else
#define flow_fault_handler FLOW_FAULT_HANDLER
#endif
#define flow_div_by_zero_handler() flow_fault_handler("division by zero")
#define flow_shift_ub_handler() flow_fault_handler("invalid shift (amount out of range or left-shift of negative)")

#ifndef FLOW_CHECKED_DIV
#define FLOW_CHECKED_DIV(L, R) (((R) != 0) ? ((L) / (R)) : (flow_div_by_zero_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_MOD
#define FLOW_CHECKED_MOD(L, R) (((R) != 0) ? ((L) % (R)) : (flow_div_by_zero_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_SHL
#define FLOW_CHECKED_SHL(L, R) ((((R) >= 0) && ((unsigned long long)(R) < (sizeof(L) * 8ull)) && ((L) >= 0)) ? ((L) << (R)) : (flow_shift_ub_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_SHR
#define FLOW_CHECKED_SHR(L, R) ((((R) >= 0) && ((unsigned long long)(R) < (sizeof(L) * 8ull))) ? ((L) >> (R)) : (flow_shift_ub_handler(), (L) * 0))
#endif

#include <math.h>

void* _ui_state = NULL;

static inline float i32_to_f32(int32_t v) { return (float)v; }

/* Host stub for @gpu kernels (device codegen replaces this). */
static inline int32_t gpu_thread_id(void) { return 0; }

int32_t main(void);

int32_t main(void) {
    int64_t fact[10] = { 1, 1, 2, 6, 24, 120, 720, 5040, 40320, 362880 };
    int64_t total = 0;
    int64_t n = 3;
    while (n <= 2540160) {
        int64_t s = 0;
        int64_t x = n;
        while (x > 0) {
            s = (s + (((unsigned)(((int32_t)(FLOW_CHECKED_MOD((x), (10))))) < 10) ? fact[((int32_t)(FLOW_CHECKED_MOD((x), (10))))] : (fprintf(stderr, "array index %d out of bounds (size %d)\n", (int)(((int32_t)(FLOW_CHECKED_MOD((x), (10))))), 10), flow_fault_handler("array index out of bounds"), fact[0])));
            x = FLOW_CHECKED_DIV((x), (10));
        }
        if (s == n) {
            total = (total + n);
        }
        n = (n + 1);
    }
    printf("%lld\n", total);
    return 0;
}

Generated MLIR

module {
  llvm.func @printf(!llvm.ptr, ...) -> i32
  llvm.mlir.global internal constant @str_0("%lld\n\00") {addr_space = 0 : i32} : !llvm.array<6 x i8>
  func.func @main() -> i32 {
    %1 = arith.constant 1 : i32
    %2 = arith.constant 1 : i32
    %3 = arith.constant 2 : i32
    %4 = arith.constant 6 : i32
    %5 = arith.constant 24 : i32
    %6 = arith.constant 120 : i32
    %7 = arith.constant 720 : i32
    %8 = arith.constant 5040 : i32
    %9 = arith.constant 40320 : i32
    %10 = arith.constant 362880 : i32
    %11 = llvm.mlir.constant(1 : i64) : i64
    %12 = llvm.alloca %11 x !llvm.array<10 x i64> : (i64) -> !llvm.ptr
    %13 = llvm.mlir.zero : !llvm.array<10 x i64>
    llvm.store %13, %12 : !llvm.array<10 x i64>, !llvm.ptr
    %14 = arith.extsi %1 : i32 to i64
    %15 = arith.extsi %2 : i32 to i64
    %16 = arith.extsi %3 : i32 to i64
    %17 = arith.extsi %4 : i32 to i64
    %18 = arith.extsi %5 : i32 to i64
    %19 = arith.extsi %6 : i32 to i64
    %20 = arith.extsi %7 : i32 to i64
    %21 = arith.extsi %8 : i32 to i64
    %22 = arith.extsi %9 : i32 to i64
    %23 = arith.extsi %10 : i32 to i64
    %24 = llvm.mlir.constant(0 : i64) : i64
    %25 = llvm.getelementptr %12[0, %24] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %14, %25 : i64, !llvm.ptr
    %26 = llvm.mlir.constant(1 : i64) : i64
    %27 = llvm.getelementptr %12[0, %26] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %15, %27 : i64, !llvm.ptr
    %28 = llvm.mlir.constant(2 : i64) : i64
    %29 = llvm.getelementptr %12[0, %28] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %16, %29 : i64, !llvm.ptr
    %30 = llvm.mlir.constant(3 : i64) : i64
    %31 = llvm.getelementptr %12[0, %30] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %17, %31 : i64, !llvm.ptr
    %32 = llvm.mlir.constant(4 : i64) : i64
    %33 = llvm.getelementptr %12[0, %32] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %18, %33 : i64, !llvm.ptr
    %34 = llvm.mlir.constant(5 : i64) : i64
    %35 = llvm.getelementptr %12[0, %34] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %19, %35 : i64, !llvm.ptr
    %36 = llvm.mlir.constant(6 : i64) : i64
    %37 = llvm.getelementptr %12[0, %36] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %20, %37 : i64, !llvm.ptr
    %38 = llvm.mlir.constant(7 : i64) : i64
    %39 = llvm.getelementptr %12[0, %38] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %21, %39 : i64, !llvm.ptr
    %40 = llvm.mlir.constant(8 : i64) : i64
    %41 = llvm.getelementptr %12[0, %40] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %22, %41 : i64, !llvm.ptr
    %42 = llvm.mlir.constant(9 : i64) : i64
    %43 = llvm.getelementptr %12[0, %42] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
    llvm.store %23, %43 : i64, !llvm.ptr
    %44 = arith.constant 0 : i32
    %45 = arith.extsi %44 : i32 to i64
    %46 = llvm.mlir.constant(1 : i64) : i64
    %47 = llvm.alloca %46 x i64 : (i64) -> !llvm.ptr
    llvm.store %45, %47 : i64, !llvm.ptr
    %48 = arith.constant 3 : i32
    %49 = arith.extsi %48 : i32 to i64
    %50 = llvm.mlir.constant(1 : i64) : i64
    %51 = llvm.alloca %50 x i64 : (i64) -> !llvm.ptr
    llvm.store %49, %51 : i64, !llvm.ptr
    cf.br ^bb0
    ^bb0:
    %52 = llvm.load %51 : !llvm.ptr -> i64
    %53 = arith.constant 2540160 : i32
    %55 = arith.extsi %53 : i32 to i64
    %54 = arith.cmpi sle, %52, %55 : i64
    cf.cond_br %54, ^bb1, ^bb2
    ^bb1:
      %56 = arith.constant 0 : i32
      %57 = arith.extsi %56 : i32 to i64
      %58 = llvm.mlir.constant(1 : i64) : i64
      %59 = llvm.alloca %58 x i64 : (i64) -> !llvm.ptr
      llvm.store %57, %59 : i64, !llvm.ptr
      %60 = llvm.load %51 : !llvm.ptr -> i64
      %61 = llvm.mlir.constant(1 : i64) : i64
      %62 = llvm.alloca %61 x i64 : (i64) -> !llvm.ptr
      llvm.store %60, %62 : i64, !llvm.ptr
      cf.br ^bb3
      ^bb3:
      %63 = llvm.load %62 : !llvm.ptr -> i64
      %64 = arith.constant 0 : i32
      %66 = arith.extsi %64 : i32 to i64
      %65 = arith.cmpi sgt, %63, %66 : i64
      cf.cond_br %65, ^bb4, ^bb5
      ^bb4:
        %67 = llvm.load %59 : !llvm.ptr -> i64
        %69 = llvm.load %62 : !llvm.ptr -> i64
        %70 = arith.constant 10 : i32
        %72 = arith.extsi %70 : i32 to i64
        %71 = arith.remsi %69, %72 : i64
        %73 = arith.trunci %71 : i64 to i32
        %74 = arith.extsi %73 : i32 to i64
        %75 = llvm.getelementptr %12[0, %74] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i64>
        %68 = llvm.load %75 : !llvm.ptr -> i64
        %76 = arith.addi %67, %68 : i64
        llvm.store %76, %59 : i64, !llvm.ptr
        %77 = llvm.load %62 : !llvm.ptr -> i64
        %78 = arith.constant 10 : i32
        %80 = arith.extsi %78 : i32 to i64
        %79 = arith.divsi %77, %80 : i64
        llvm.store %79, %62 : i64, !llvm.ptr
        cf.br ^bb3
      ^bb5:
      %81 = llvm.load %59 : !llvm.ptr -> i64
      %82 = llvm.load %51 : !llvm.ptr -> i64
      %83 = arith.cmpi eq, %81, %82 : i64
      cf.cond_br %83, ^bb6, ^bb7
      ^bb6:
        %84 = llvm.load %47 : !llvm.ptr -> i64
        %85 = llvm.load %51 : !llvm.ptr -> i64
        %86 = arith.addi %84, %85 : i64
        llvm.store %86, %47 : i64, !llvm.ptr
        cf.br ^bb8
      ^bb7:
        cf.br ^bb8
      ^bb8:
      %87 = llvm.load %51 : !llvm.ptr -> i64
      %88 = arith.constant 1 : i32
      %90 = arith.extsi %88 : i32 to i64
      %89 = arith.addi %87, %90 : i64
      llvm.store %89, %51 : i64, !llvm.ptr
      cf.br ^bb0
    ^bb2:
    %91 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %92 = llvm.load %47 : !llvm.ptr -> i64
    %93 = llvm.call @printf(%91, %92) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %94 = arith.constant 0 : i32
    func.return %94 : i32
  }
}